Update 12/06/2024 by Martijn van Best – French AI startup Mistral AI has raised the desired 600 million euros it previously set as the target amount for its latest investment round. Its estimated enterprise value is now close to €6 billion.
This latest funding from Mistral comprises €468 million in equity and €132 million in debt, reports the Financial Times. The company, which is barely a year old, is now valued 5.8 billion euros.
This time, the main backer is venture capital firm General Catalyst, along with a few other current investors who have made additional contributions. These are investment firms Lightspeed and Andreessen Horowitz, as well as banks Bpifrance and BNP Paribas. Corporates such as Nvidia, Samsung, IBM, and Salesforce also participate. Microsoft announced in February investing over 15 million euros in the French AI startup.
Founders Timothée Lacroix, Guillaume Lample and Arthur Mensch remain majority shareholders.
Original article 10/5/2024 – Mistral AI wants to raise $600 million with an estimated valuation of $6 billion in a new round of investment.
The AI developer is in talks for a new investment round, writes The Wall Street Journal. The targeted financial boost is significantly higher than what we already reported on last month. Back then, Mistral hoped to find $500 million with a market value of $5 billion.
The last investment round last December raised $415 million. This puts the estimated value of France’s Mistral at $2 billion.
Greater confidence in own ability
The raising of the investment amount may be because Mistral AI was more confident about the company’s growth over the past month. This is said to be prompted by the increased popularity of its own LLMs, which are available on an increasing number of platforms. It would also require the new valuation to be commensurate with the AI innovations it has produced recently.
For example, Mistral recently introduced the AI model Mixtral 8x22B. This LLM scored very well in various benchmarks and can compete with well-known LLMs from competitors such as OpenAI and Google. For example, it would already come close to the capacity of Meta’s brand-new Llama 3.
Use new investment
The money raised by the new round of investment can obviously be used by AI developers to develop new, even more powerful LLMs. One possibility is to focus on targeted models for text, video and sound, for example.
However, Mistral is also thinking beyond LLMs. The company is already betting on this with Mistral Embed. This is an AI model designed to turn text into embeddings. These are mathematical structures that neural networks can process and is a necessary step to get GenAI up and running. Mistral AI’s tool should become a competitor to the current open-source fastText that also generates embeddings.
Also read: Mistral introduces new powerful Mixtral 8x22B LLM